You can also use a simple Brightness index using visible bands and having object-oriented approach. Then compare your results with the method developed in the above paper and other indices like MSAVI, TSAVI, etc.
Visible bands (B1, B2 B3 for ETM+) are used to compute brightness index. For other indices like MSAVI, red band and NIR band (B3 and B4) are commonly used.
Do you work on an area containing water (wetland or coastal sites) or different soils? In such case, you should use an index taking into account these parameters like TSAVI.
In urban area you might have confusion between soil and roofs or roads. I am not an expert in urban RS. May be you will find an answer in this paper. Landsat-8 and ETM+ are very similar.
This is a common question which many people ask as rightly pointed out by USGS. Check attached link.
Atiqa, I think your question has many loose ends. First, what kind of soil are we talking about? Is it a sand dominated soil, clay, loam etc.? What is the percentage of Soil cover in the whole Landsat Scene? Even if you know the soil type, it depends on how abundant the soil is and how homogeneously it is spread across your study area. You also later on mention that your study area is densely populated? Can you name the area specifically?
Given the 30m pixel size of ETM+, unless and until the whole pixel is covered by one soil type, you will generally not find the expected spectral signature of the soil because in a 30 by 30 m pixel the spectral signature you get is a combination of various other sub-pixel land covers contributed by vegetation, urban features, barren and soil. Therefore in ETM+ pixels the best way to find the suitable bands depends on the abundance of the soil type, homogeneity etc. Also check out this useful spectral Characteristics Viewer from USGS. Links are attached.
Actually, it s Lahore, Pakistan... Urban soil, im talking about, if so exist as in small patches in this highly populated city. Me working with Landsat 7... And i want a generalized overview to be used in further understanding.